To Scan or Not to Scan: Personalizing Lung Cancer Screening
A new tool that takes into account personalized risks and benefits, and allows for a range of patient preferences, will help clinicians decide whom to screen for lung cancer.
For many smokers and former smokers, the risk of someday developing lung cancer is high.
To catch potential problems early, some decide to have a CT scan of their lungs, which can find a tumor early enough to stop it — or set off a false alarm.
Others may avoid the scans, or don’t know they should have one, even though they’re considered at high risk under the official recommendations in effect for the past five years.
To help, a new study shows how to personalize the lung cancer screening decision for every patient. The results could help doctors fine-tune their advice to patients so that it’s based not only on a patient’s individual lung cancer risk and the potential benefits and harms of screening, but also the patient’s attitude about looking for problems and dealing with the consequences.
Published in the Annals of Internal Medicine, the study forms the backbone for free new online tools: one for health care providers, one for members of the public.
The tool for clinicians, called Lung Decision Precision, was designed by a University of Michigan and Veterans Affairs team to help clinicians talk with patients and their loved ones about lung CT scans.
The patient-focused website, ShouldIScreen.com, gives easy-to-understand information about the positives and potential negatives of lung cancer screening, and it allows individuals to calculate their personal risk of lung cancer.
Turning data into actions
Tanner Caverly, M.D., MPH, led the team that created the models using simulation analysis of data from major studies of lung cancer screening, and national data on the potential screening population under the current guidelines.
“Our model is built on a comprehensive view of net benefits for individual patients, which incorporates the best evidence for personalizing the pros and cons of screening, and assumes that not all patients will feel the same about screening and its consequences,” says Caverly, an assistant professor in the Division of General Medicine and Department of Learning Health Sciences at the U-M Medical School.
“This allows us to identify which patients are in the preference-sensitive zone for the decision about screening and which ones have a very clear potential benefit to them.”
Any person with an annual chance of lung cancer between 0.3 percent and 1.3 percent and a life expectancy of more than 10 years falls into the high-benefit category, he notes. This accounts for about 50 percent of all Americans who qualify for screening under the current guidelines.
But for most of the rest, personal preferences should help determine if they should get screened. For example, how much they dislike medical testing in general, how they feel about the potential unintended consequences of looking for a problem when they feel fine, and how they view the process of follow-up scans and lung biopsies if the initial screen shows something suspicious.
In fact, for such patients, personal preference is more important to their decision than the false-positive rate for lung CT screening (which outnumber true cancers 25 to 1), and the negative effects of overtreatment for a lung cancer that was not highly dangerous.
At the same time, people who fall into the potential pool of screening candidates but have a short life expectancy and a low risk of lung cancer should probably not be screened, the researchers say.
The model could help physicians prepare for conversations with patients about lung screening, customized per person.
“If a physician is not clear about the potential benefit for a patient who’s in the high-benefit zone, they could miss an opportunity to do something really good for them, to say, ‘I don’t recommend this for everyone but I recommend it for you,’” Caverly says.
“But coming across strong for screening with a patient who has a fine balance of pros and cons could miss an opportunity to give them a choice, to tell them that their decision depends on their personality.”
While the study looked specifically at the evidence around lung cancer screening, the authors note the underlying analytical method could lead to personalized health decision tools for other situations.
Looking at lungs
The researchers focused on lung cancer — the leading cause of cancer death among both men and women — because of the recent move to encourage certain smokers and former smokers between the ages of 55 and 80 to get screened for it.
A major study published in 2011 showed that some members of this group could survive longer if they had CT screening to find the earliest signs of lung cancer, which is diagnosed in more than 230,000 Americans every year.
Two years later, a national panel recommended it for people between the ages of 55 and 80 who had been or still were heavy smokers, defined as an average of a pack of cigarettes a day for 30 years for people who currently smoke or quit less than 15 years ago.
The recommendation comes with exceptions, even among this group, including patients whose overall health meant they have a life expectancy of less than 10 years, and those who aren’t strong enough to withstand lung surgery if a scan shows signs of cancer.
Now, the U.S. Preventive Services Task Force, which made the initial recommendation, is preparing to revisit its guidance on this topic.
Caverly and his colleagues hope their new study will inform that process. Among those involved in the study was senior author Rafael Meza, Ph.D., who is coordinating principal investigator of the Cancer Intervention and Surveillance Modeling Network (CISNET) lung group that did the decision modeling supporting the current Task Force recommendations. Meza is an associate professor in the U-M School of Public Health.
They’re studying how the online tool can be used by physicians and the medical team members who assist with screening. They’re also thinking about whether it could be included in the online systems that patients use to communicate with their clinic ahead of an appointment, and the clinical reminder tools that prompt physicians to talk with patients about preventive services that might be right for them.
The physician-focused tool produces a colorful display that places the individual patient somewhere along a green and yellow line. If a patient is deep in the yellow, they likely have a small but non-zero benefit with screening and screening will depend highly on patient views.
The closer they fall to the divide between green and yellow, the more likely it is that screening will benefit them. And if they’re deep in the green zone, the physician should encourage them more strongly to get screened. The site generates other visual representations and handouts for patients and their spouse or other loved one.
The team, including co-author Rodney Hayward, M.D., of General Medicine and the VA Ann Arbor Healthcare System, and Pianpian Cao, doctoral student in epidemiology at U-M, hopes to test the model’s usefulness for other types of health services that are used by many people and have good data available about their benefits, risks and patient preferences.
This includes cancer screening, disease prevention, chronic disease management approaches and more.
“This method can incorporate anything that moves the needle on risk and benefit, and that involves patient preferences about time, dollars and worry,” Caverly says. “As a clinician I’d like to have this for many of the things I do, where it would be meaningful to know how beneficial something could be for the individual patient, and we could talk about whether it’s indicated for them.”
Caverly, Meza and Hayward are members of the U-M Institute for Healthcare Policy and Innovation. Caverly and Hayward are members of the VA Center for Clinical Management research and received VA funding to develop the web-based tool for providers. Meza and Cao hold funding from the National Cancer Institute; Meza co-leads the Cancer Epidemiology and Prevention Program at the U-M Rogel Cancer Center.